Debian Bug report logs - #765694
hardcoded libeatmydata path now is wrong

Toggle useless messages

Message #5 received at submit@bugs.debian.org (full text, mbox, reply):

From: Mattia Rizzolo <mattia@mapreri.org>

To: Debian Bug Tracking System <submit@bugs.debian.org>

Subject: hardcoded libeatmydata path now is wrong

Date: Fri, 17 Oct 2014 13:57:09 +0200

[Message part 1 (text/plain, inline)]

Package: ltsp
Version: 5.5.3-1

The new upload of libeatmydata changed the position of the actual library (due
to now supporting Multi-Arch).

You appeat to be using /usr/lib/libeatmydata/libeatmydata.so directly:
http://sources.debian.net/src/ltsp/5.5.3-1/server/share/ltsp/plugins/ltsp-build-client/Debian/080-eatmydata/?hl=8#L8

I don't know what's the purpose of it (given that you don't
depends/recommends/suggest eatmydata), but for sure now it'll do just nothing.
Please set the severity of this bug accordingly.

I'd suggest you to just use "libeatmydata.so" instead of using the full path.

-- 
regards,
						Mattia Rizzolo

GPG Key: 4096R/B9444540	http://goo.gl/I8TMB
more about me:		http://mapreri.org
Launchpad User:		https://launchpad.net/~mapreri
Ubuntu Wiki page:	https://wiki.ubuntu.com/MattiaRizzolo

[signature.asc (application/pgp-signature, inline)]

Message #10 received at 765694@bugs.debian.org (full text, mbox, reply):

From: Vagrant Cascadian <vagrant@debian.org>

To: Mattia Rizzolo <mattia@mapreri.org>, 765694@bugs.debian.org

Subject: Re: Bug#765694: hardcoded libeatmydata path now is wrong

Date: Fri, 17 Oct 2014 12:22:28 -0700

[Message part 1 (text/plain, inline)]

Control: tags -1 pending fixed-upstream

On 2014-10-17, Mattia Rizzolo wrote:
> The new upload of libeatmydata changed the position of the actual library (due
> to now supporting Multi-Arch).
...
> I'd suggest you to just use "libeatmydata.so" instead of using the full path.

Thanks for the bug report!

Will upload a fix soon, committed upstream:

  https://bazaar.launchpad.net/~ltsp-upstream/ltsp/ltsp-trunk/revision/2609


live well,
  vagrant

[Message part 2 (application/pgp-signature, inline)]

Message #17 received at 765694@bugs.debian.org (full text, mbox, reply):

From: Petter Reinholdtsen <pere@hungry.com>

To: 765694@bugs.debian.org

Cc: Mattia Rizzolo <mattia@mapreri.org>

Subject: Re: [Pkg-ltsp-devel] Bug#765694: hardcoded libeatmydata path now is wrong

Date: Fri, 17 Oct 2014 22:58:10 +0200

[Vagrant Cascadian]
> Will upload a fix soon, committed upstream:
> 
>   https://bazaar.launchpad.net/~ltsp-upstream/ltsp/ltsp-trunk/revision/2609

Great.  Btw, do we need to do something extra to get this enabled in
Debian Edu, or is it the default?  We install the rest of the system
using eatmydata now, and it would be great to speed up the LTSP chroot
building if possible.

-- 
Happy hacking
Petter Reinholdtsen

Message #22 received at 765694@bugs.debian.org (full text, mbox, reply):

From: Vagrant Cascadian <vagrant@debian.org>

To: Petter Reinholdtsen <pere@hungry.com>, 765694@bugs.debian.org

Subject: Re: Bug#765694: hardcoded libeatmydata path now is wrong

Date: Fri, 17 Oct 2014 14:38:00 -0700

[Message part 1 (text/plain, inline)]

On 2014-10-17, Petter Reinholdtsen wrote:
> Great.  Btw, do we need to do something extra to get this enabled in
> Debian Edu, or is it the default?  We install the rest of the system
> using eatmydata now, and it would be great to speed up the LTSP chroot
> building if possible.

You need to pass ltsp-build-client the --eatmydata flag; it is not
enabled by default. This feature was introduced in LTSP in 2011...

live well,
  vagrant

[Message part 2 (application/pgp-signature, inline)]

Message #27 received at 765694-close@bugs.debian.org (full text, mbox, reply):

From: Vagrant Cascadian <vagrant@debian.org>

To: 765694-close@bugs.debian.org

Subject: Bug#765694: fixed in ltsp 5.5.4-1

Date: Wed, 22 Oct 2014 07:04:54 +0000

Source: ltsp
Source-Version: 5.5.4-1

We believe that the bug you reported is fixed in the latest version of
ltsp, which is due to be installed in the Debian FTP archive.

A summary of the changes between this version and the previous one is
attached.

Thank you for reporting the bug, which will now be closed.  If you
have further comments please address them to 765694@bugs.debian.org,
and the maintainer will reopen the bug report if appropriate.

Debian distribution maintenance software
pp.
Vagrant Cascadian <vagrant@debian.org> (supplier of updated ltsp package)

(This message was generated automatically at their request; if you
believe that there is a problem with it please contact the archive
administrators by mailing ftpmaster@ftp-master.debian.org)


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Format: 1.8
Date: Tue, 21 Oct 2014 23:32:58 -0700
Source: ltsp
Binary: ltsp-server ltsp-server-standalone ltsp-client-core ltsp-client ltsp-client-builder
Architecture: source all
Version: 5.5.4-1
Distribution: unstable
Urgency: medium
Maintainer: LTSP Debian Maintainers <pkg-ltsp-devel@lists.alioth.debian.org>
Changed-By: Vagrant Cascadian <vagrant@debian.org>
Description:
 ltsp-client - complete LTSP client environment
 ltsp-client-builder - build an LTSP environment in the installer target (udeb)
 ltsp-client-core - basic LTSP client environment
 ltsp-server - basic LTSP server environment
 ltsp-server-standalone - complete LTSP server environment
Closes: 761891 765065 765443 765694
Changes:
 ltsp (5.5.4-1) unstable; urgency=medium
 .
   * New upstream version:
     - Support for disabling systemd services with RM_SYSTEM_SERVICES.
       (Closes: #765065)
 .
     - Disable cups-browsed and ModemManager services by default.
       (Closes: #761891, #765065).
 .
     - Set up hostname.domain in /etc/hosts (LP: #1093088).
 .
     - Fix use of SHUTDOWN_TIME with LDM to use ltsp-logout-action instead.
       (LP: #604908).
 .
     - Fix ssh, xdmcp and telnet screen scripts by checking for empty
       argument (LP: #1020113, #1175618).
 .
     - Default to "server" in ssh screen script so that ssh_known_hosts
       file generated by ltsp-update-sshkeys will match.
 .
     - ltsp-build-client: Do not use full path to libeatmydata.so.
       (Closes: #765694)
 .
     - Unset temporary directory variables to avoid problems when directory
       is missing inside the chroot (Closes: #765443).
 .
     - update-kernels: Update ARM code to generate a generic bootscript.
 .
   * debian/control:
     - Add u-boot-tools to ltsp-client-core Depends on armhf/armel.
     - Update Standards-Version to 3.9.6, no changes needed.
Checksums-Sha1:
 c8d0fd4ed31ee1dc89d7f951a3680362689b07c7 2285 ltsp_5.5.4-1.dsc
 a1cebda26315ebd3d3dbbc16753a6e60cb349fec 311424 ltsp_5.5.4.orig.tar.xz
 e60ea122eaa866b49876063cb30b5cda15cd8625 73860 ltsp_5.5.4-1.debian.tar.xz
 7c8c57c7704bb25bc5cfddcfab0396e817fd77fb 266194 ltsp-server_5.5.4-1_all.deb
 5d7c29217dad4155abf746ecf5372a67be592b4e 145910 ltsp-server-standalone_5.5.4-1_all.deb
 92ff52f2a81b26c5e1be9c9f5565bb591d2b6d06 11124 ltsp-client-builder_5.5.4-1_all.udeb
Checksums-Sha256:
 1ba43a010e7f6e94e492fc3a1cb944f9aebd23ef69b7373c500025d1fa0bb0d5 2285 ltsp_5.5.4-1.dsc
 48104820c6beecf6017d5a287521ea32239179a78a30361af77b0bef72a728f3 311424 ltsp_5.5.4.orig.tar.xz
 bbca8fbaef2b3a7c48200a6f796507589096aef2179e022f5e586e4b6e540dcd 73860 ltsp_5.5.4-1.debian.tar.xz
 7e003538dd627e1a919466bb336f72c7a6e74788c2419ef5e560b8c8d4808c6a 266194 ltsp-server_5.5.4-1_all.deb
 ff712bf03387c6e061d68c040eacdc3179a6abf2b89304551ef66e8bb1a7cbb3 145910 ltsp-server-standalone_5.5.4-1_all.deb
 db8e27f48355443a7c8a6dc03ee3e216103cc1431ffc4a1a5d15efe9d4612bf3 11124 ltsp-client-builder_5.5.4-1_all.udeb
Files:
 78fa9f6d4ab028ae6a3ccf6da47778ad 2285 misc extra ltsp_5.5.4-1.dsc
 e7e37c2cb1b471e09e31280d9c4f85cb 311424 misc extra ltsp_5.5.4.orig.tar.xz
 471fb08f52ad33b98f670232988cbec5 73860 misc extra ltsp_5.5.4-1.debian.tar.xz
 dd2bd33f28be7c10372e1e32481ef850 266194 misc extra ltsp-server_5.5.4-1_all.deb
 fbf87330a329f0ecd6c75338dce562aa 145910 misc extra ltsp-server-standalone_5.5.4-1_all.deb
 063d87b89edb652455e9b034b794109d 11124 debian-installer extra ltsp-client-builder_5.5.4-1_all.udeb

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
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=BZBw
-----END PGP SIGNATURE-----

Message #32 received at 765694@bugs.debian.org (full text, mbox, reply):

From: Petter Reinholdtsen <pere@hungry.com>

To: 765694@bugs.debian.org

Subject: Re: Bug#765694: hardcoded libeatmydata path now is wrong

Date: Mon, 27 Oct 2014 12:05:47 +0100

[Vagrant Cascadian]
> You need to pass ltsp-build-client the --eatmydata flag; it is not
> enabled by default. This feature was introduced in LTSP in 2011...

We do this now in Debian Edu, but there is something wrong with the
new ltsp version.  I see lots of "ERROR: ld.so: object
'libeatmydata.so' from LD_PRELOAD cannot be preload (cannot open
shared object file): ignored" messages in the installation log.

The strange thing is that later in the installation, exim4-config
trigger #762103, which is supposed to be fixed in the newer eatmydata
package entering testing yesterday.

Is --eatmydata working now for you when using Jessie?  It isn't for me
any more.

-- 
Happy hacking
Petter Reinholdtsen

Message #37 received at 765694@bugs.debian.org (full text, mbox, reply):

From: Mattia Rizzolo <mattia@mapreri.org>

To: Petter Reinholdtsen <pere@hungry.com>, 765694@bugs.debian.org

Subject: Re: Bug#765694: hardcoded libeatmydata path now is wrong

Date: Mon, 27 Oct 2014 15:20:54 +0100

On Mon, Oct 27, 2014 at 12:05 PM, Petter Reinholdtsen <pere@hungry.com> wrote:
> The strange thing is that later in the installation, exim4-config
> trigger #762103, which is supposed to be fixed in the newer eatmydata
> package entering testing yesterday.

Not related to what you are reporting in this bug, but this can be
caused by https://bugs.debian.org/765738 which is a really odd bug,
and  furthermore reproducible only on i386 (or 32 bit arches, not sure
here). I'm currently investigating this.

-- 
regards,
Mattia Rizzolo

GPG Key: 4096R/B9444540 http://goo.gl/I8TMB
more about me: http://mapreri.org
Launchpad User: https://launchpad.net/~mapreri
Ubuntu Wiki page: https://wiki.ubuntu.com/MattiaRizzolo

Message #42 received at 765694@bugs.debian.org (full text, mbox, reply):

From: Petter Reinholdtsen <pere@hungry.com>

To: Mattia Rizzolo <mattia@mapreri.org>, 765694@bugs.debian.org

Subject: Re: Bug#765694: hardcoded libeatmydata path now is wrong

Date: Mon, 27 Oct 2014 15:32:50 +0100

[Mattia Rizzolo]
> Not related to what you are reporting in this bug, but this can be
> caused by https://bugs.debian.org/765738 which is a really odd bug,
> and furthermore reproducible only on i386 (or 32 bit arches, not
> sure here). I'm currently investigating this.

Right.  The LTSP chroot in Debian Edu is i386, so that sound likely.
It is the same messages as #762103 triggered by bug #702711 which was
fixed in eatmydata version 26-2.1.

This explain why we did not see the problem with LTSP when using
26-2.1, and also why installing amd64 machines work but not the i386
chroot.

-- 
Happy hacking
Petter Reinholdtsen

Message #47 received at 765694@bugs.debian.org (full text, mbox, reply):

From: Vagrant Cascadian <vagrant@debian.org>

To: Petter Reinholdtsen <pere@hungry.com>, 765694@bugs.debian.org, 765694-submitter@bugs.debian.org

Subject: Re: Bug#765694: hardcoded libeatmydata path now is wrong

Date: Mon, 27 Oct 2014 10:21:25 -0700

[Message part 1 (text/plain, inline)]

On 2014-10-27, Petter Reinholdtsen wrote:
> We do this now in Debian Edu, but there is something wrong with the
> new ltsp version.  I see lots of "ERROR: ld.so: object
> 'libeatmydata.so' from LD_PRELOAD cannot be preload (cannot open
> shared object file): ignored" messages in the installation log.
>
> The strange thing is that later in the installation, exim4-config
> trigger #762103, which is supposed to be fixed in the newer eatmydata
> package entering testing yesterday.
>
> Is --eatmydata working now for you when using Jessie?  It isn't for me
> any more.

I'll try again, now that eatmydata has migrated to jessie.

It worked with amd64 server with and amd64 chroot with newer versions in
all places.

I had troubles with an amd64 server with an i386 chroot, with the newer
version of libeatmydata.

It didn't work with older versions of eatmydata at all, which is
unfortunate.

So I've had mixed results.

It dramatically improves speed building LTSP chroots, so I'd really like
to get it working more reliably, but figured even in the imperfect state
it was better to move forward where it works in some cases than leave it
totally broken.

live well,
  vagrant

[Message part 2 (application/pgp-signature, inline)]

Message #55 received at 765694@bugs.debian.org (full text, mbox, reply):

From: Petter Reinholdtsen <pere@hungry.com>

To: 765694@bugs.debian.org, 765694-submitter@bugs.debian.org

Subject: Re: Bug#765694: hardcoded libeatmydata path now is wrong

Date: Tue, 11 Nov 2014 19:04:36 +0100

[Vagrant Cascadian]
> I had troubles with an amd64 server with an i386 chroot, with the
> newer version of libeatmydata.

I believe this is due to <URL: http://bugs.debian.org/702711 > causing
eatmydata to fail on i386.  It is fixed in unstable and a fix should
hopefully make it into testing if the libeatmydata unblock request,
<URL: http://bugs.debian.org/768893 >, is accpted.

-- 
Happy hacking
Petter Reinholdtsen

Send a report that this bug log contains spam.